PeterSamuelR/ML-based-cloud-resource-prediction-and-auto-scaling-system ? reverse-engineered prompt

Build me a reproducible local demo of a machine learning based cloud autoscaling system.

I want it to run with Docker Compose and include a small web app behind Nginx, a backend that watches usage every few seconds, stores metrics, predicts future CPU load with a model, and decides whether to scale up or down. Please include three modes, reactive scaling, predictive scaling, and an adaptive predictive mode that learns from prediction error and can replace a model when the new one is better.

Also make a simple dashboard that shows the current system state, charts for usage and predictions, scaling history, and experiment results, with live updates from the backend. Add a workload generator so I can test it locally, and make the whole thing easy to start and stop with clear defaults and a sample config. If you need to check current docs or best practices for any libraries, look them up online.